TechJul 1, 2026
Meta's Next Act Is Taking On the Cloud Giants
Infrastructure Pivot
Rhea Lobo
Discover
Meta Platforms jumped on July 1 after Bloomberg reported the company is building a cloud business to sell excess AI computing capacity to outside customers.
The initiative, called Meta Compute, is led by Meta's head of infrastructure Santosh Janardhan alongside Daniel Gross from Meta's Superintelligence Labs and Meta President Dina Powell McCormick.
Two products are under consideration. The first would let developers access AI models hosted on Meta's infrastructure, including its Muse Spark models, and pay for the computing power used. The second would sell raw GPU capacity directly, similar to what neocloud companies currently offer.
Meta has told investors it plans to spend as much as $145B on capital expenditures this year. That figure has made shareholders uneasy about whether the spending will ever generate a return.
A cloud business offers a direct answer to that concern. Amazon, Microsoft, and Alphabet have each built cloud divisions that now generate tens of billions of dollars per quarter. Meta has historically used its infrastructure entirely for internal workloads, so entering commercial cloud services marks a major strategic shift.
CEO Mark Zuckerberg had signaled this possibility at the company's May shareholder meeting.
"Almost every week there are different companies that come to us from the outside asking us to both stand up an API service or asking if we have compute that they could buy from us at some premium to what we've bought it at."
Mark Zuckerberg, Meta
He added that Meta hadn't moved yet because it still had use for the compute, but that selling excess capacity was "definitely on the table" if the company felt it had overbuilt.
Meta's move could increase competition in AI cloud infrastructure, particularly for specialized GPU cloud providers.
That prompted a selloff in CoreWeave and Nebius Group as both companies count Meta as a major customer. Nebius holds a deal with Meta worth up to ~$27B over five years. CoreWeave's agreement with Meta totals ~$21B.
If Meta builds out its own external cloud business, it may no longer need to buy as much capacity from those providers. It could also start competing directly with them for the same customers.
Meta isn't the first company to take this path. Bloomberg reported that SpaceX has started leasing capacity from its Memphis data center to outside customers. Anthropic agreed to pay $1.25B per month for that capacity. Google agreed to pay $920M per month.
The neocloud business model was built on a simple premise: AI demand was growing faster than traditional cloud providers could respond. That created an opening for specialized GPU-focused suppliers. But as hyperscalers and large AI developers build more of their own capacity, that opening narrows.
Analysts note that customer concentration is the structural risk. Companies whose revenue depends heavily on a single large customer are exposed when that customer decides to go in a different direction. Meta's move doesn't signal that AI compute demand is shrinking. It signals that the companies supplying that demand are changing.
